Do not pre-migrate schema1 docker tags when there are 2 tags with sam…

…e name witin a repo.

closes #6234
https://pulp.plan.io/issues/6234

CHANGES/6234.bugfix

@@ -0,0 +1 @@
+Do not pre-migrate schema1 docker tags when there are 2 tags with same name witin a repo.

pulp_2to3_migration/app/plugin/docker/migrator.py

@@ -5,6 +5,7 @@
 from django.db import IntegrityError
 
 from . import pulp2_models
+from . import utils
 
 from .pulp_2to3_models import (
     Pulp2Blob,
@@ -75,6 +76,10 @@ class DockerMigrator(Pulp2to3PluginMigrator):
         'docker_distributor_web': DockerDistributor,
     }
 
+    premigrate_hook = {
+        'docker_tag': utils.find_tags
+    }
+
     @classmethod
     async def migrate_content_to_pulp3(cls):
         """

pulp_2to3_migration/app/plugin/docker/utils.py

@@ -0,0 +1,20 @@
+from . import pulp2_models
+
+
+def find_tags():
+    """
+    Find tags that have same name within the repo.
+    Return only one tag out of 2 tags with the same name.
+    Prefer schema2 over schema1.
+    """
+
+    # sort the schema version in desc mode.
+    sort_stage = {'$sort': {'schema_version': -1}}
+    # group tags by name and repo_id; take just first result out of the 2 tags with the same name
+    group_stage1 = {'$group': {'_id': {'name': '$name', 'repo_id': '$repo_id'},
+                    'tags_id': {'$first': '$_id'}}}
+    group_stage2 = {'$group': {'_id': None, 'tags_ids': {'$addToSet': '$tags_id'}}}
+    result = pulp2_models.Tag.objects.aggregate([sort_stage, group_stage1, group_stage2])
+    if result._has_next():
+        return result.next()['tags_ids']
+    return []

pulp_2to3_migration/app/plugin/iso/migrator.py

@@ -44,6 +44,8 @@ class IsoMigrator(Pulp2to3PluginMigrator):
         'iso_distributor': IsoDistributor,
     }
 
+    premigrate_hook = {}
+
     @classmethod
     async def migrate_content_to_pulp3(cls):
         """

pulp_2to3_migration/app/pre_migration.py

@@ -55,13 +55,17 @@ async def pre_migrate_all_content(plan):
                                          pulp_2to3_detail=pulp_2to3_detail_model)
             # identify wether the content is mutable
             mutable_type = content_model.pulp2.TYPE_ID in plugin.migrator.mutable_content_models
-            pre_migrators.append(pre_migrate_content(content_model, mutable_type))
+            # check if the content type has a premigration hook
+            premigrate_hook = None
+            if content_model.pulp2.TYPE_ID in plugin.migrator.premigrate_hook:
+                premigrate_hook = plugin.migrator.premigrate_hook[content_model.pulp2.TYPE_ID]
+            pre_migrators.append(pre_migrate_content(content_model, mutable_type, premigrate_hook))
 
     _logger.debug('Pre-migrating Pulp 2 content')
     await asyncio.wait(pre_migrators)
 
 
-async def pre_migrate_content(content_model, mutable_type):
+async def pre_migrate_content(content_model, mutable_type, premigrate_hook):
     """
     A coroutine to pre-migrate Pulp 2 content, including all details for on_demand content.
 
@@ -81,8 +85,13 @@ async def pre_migrate_content(content_model, mutable_type):
         type=content_type,
         timestamp=last_updated))
 
-    # query only newly created/updated items
-    mongo_content_qs = content_model.pulp2.objects(_last_updated__gte=last_updated)
+    if premigrate_hook:
+        pulp2_content_ids = premigrate_hook()
+        mongo_content_qs = content_model.pulp2.objects(
+            _last_updated__gte=last_updated, id__in=pulp2_content_ids)
+    else:
+        # query only newly created/updated items
+        mongo_content_qs = content_model.pulp2.objects(_last_updated__gte=last_updated)
     total_content = mongo_content_qs.count()
     _logger.debug('Total count for {type} content to migrate: {total}'.format(
         type=content_type,